Counter-terrorism legislation as a citation network
A demo from a doctoral corpus of counter-terrorism legislation. Every figure below is a real
neighbourhood exported from the Ukrainian corpus — 348,914 documents, 570,066 sections,
862,525 citation edges and 25,301 amendment edges in a Neo4j graph. Arrows show direction:
a grey arrow is a citation, a green arrow is an
amendment — one act changing another.
What this is showing, and what it is not
- Amendment edges are the point. The argument this corpus supports is about
layering — new law stacked onto inherited law rather than replacing it. That is a claim
about direction, so citations and amendments are drawn as separate edge kinds and never
collapsed together.
- Anchor duplicates are excluded. Ukrainian citations often target a specific
paragraph of a law, which the source portal exposes as a separate identifier
(
361-20/paran1006) even though it is a full copy of the base act. Those are
dropped here — left in, a law appears to cite several duplicates of itself.
- Layout is deterministic. No randomness: the same law always draws the same figure,
so any of these can be cited as a stable figure.